docs: 📝 Add expected answers to DataFrame method examples

[Eason0729]

Which issue does this PR close?

Closes #12527.

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

[alamb]

What would you think about using assert_batches_eq instead so that the output is automatically checked as part of the doc tests rather than relying on us manually keeping it up to date?

It is of similar readability I think. Here is an example:

datafusion/datafusion/core/tests/sql/select.rs

Lines 27 to 49 in d9cb6e6

	let results = ctx
	.sql("SELECT * FROM test WHERE c1 = $1")
	.await?
	.with_param_values(vec![ScalarValue::from(3i32)])?
	.collect()
	.await?;
	let expected = vec![
	"+----+----+-------+",
	"| c1 | c2 | c3 |",
	"+----+----+-------+",
	"| 3 | 1 | false |",
	"| 3 | 10 | true |",
	"| 3 | 2 | true |",
	"| 3 | 3 | false |",
	"| 3 | 4 | true |",
	"| 3 | 5 | false |",
	"| 3 | 6 | true |",
	"| 3 | 7 | false |",
	"| 3 | 8 | true |",
	"| 3 | 9 | false |",
	"+----+----+-------+",
	];
	assert_batches_sorted_eq!(expected, &results);

[Eason0729]

My original thought is to simplify large long column on doctest(less readable), like this: https://github.com/Eason0729/datafusion/blob/4dd44c5a2b0d9810d7e9163689afab227c58d542/datafusion/core/src/dataframe/mod.rs#L734

Therefore, I simplify example_long.csv in the next commit to the point which is just enough to showcase most method on dataframe.

[Eason0729]

But the expect behavior on describe method is more complex, which might require dedicated csv, so I decided to leave as it is.

[alamb]

Thank you so much @Eason0729 for this contribution 🙏

I have started the CI to check this code

I think the PR would be even better if the output was verified (so it is guaranteed to stay in sync). I left a suggestion on how to do this. Let me know what you think

[comphead]

👍

[comphead]

Suggested change

	/// // expend into multiple columns if it's json array, flatten field name if it's nested structure
	/// // expand into multiple columns if it's json array, flatten field name if it's nested structure

[comphead]

Thanks @Eason0729 for your contribution.
I'm feeling we need to have something to create DF from rows rather in addition to creating DF from data files.
Example can be

let schema = 
DataFrame::from(schema, data)

Underneath the method can call ctx.read_batch(record_batch). The batch can be created with RecordBatch::try_from_iter